Evaluating mechanical systems before position sizing

Evaluation as a gate

Editorial reading: treat evaluation as the gate that decides whether a mechanical trading system is stable enough to receive any position-sizing rule, rather than as a scoreboard after the system is already being sized.

A mechanical trading system is a fully specified procedure that turns rule inputs, market state, and execution constraints into entry, exit, or abstention signals over the system's holding period. Until that procedure has been checked, a sizing rule only scales an unevaluated design.

The five-step process

The archive presents a five-step performance analysis process: code the methodology, run robustness analysis for curve-fitting, evaluate the individual system, apply system-level position sizing, then combine systems in a portfolio with portfolio-level sizing.

Robustness analysis is the check, after coding, that tries to determine whether a methodology is curve-fitted. The archive places that check before deeper evaluation and before any money-management overlay.

Walk-forward checks and optimization

Walk-forward analysis is a sequential out-of-sample evaluation method that tests whether coded rules remain usable as market conditions change. The archive treats walk-forward and related out-of-sample checks as necessary because a single historical test is not a fixed forecast of future system behavior.

System optimization, in this workflow, means adjusting a system's rules or parameters only after robustness and evaluation checks show the design is not merely fitted to one historical path.

Sizing after evaluation

Position sizing is a separate algorithm that changes how many contracts or shares a system trades as its environment changes, rather than rewriting the system's entry and exit rules.

System-level money management is described as using approaches such as maximum adverse excursion or maximum favorable excursion to change position size rather than a constant share or contract count. Maximum favorable excursion is associated with trend-oriented systems that scale exposure from favorable price movement.

The archive argues that matching a sizing method to a system's behavior and to the trader matters, because an aggressive sizing algorithm on an erratic system can make the whole plan fail.

The archive also states that systems can remain useful in an unfriendly regime if position size is reduced, instead of rewriting trend rules to avoid consolidation.

Portfolios and a normalized display

Portfolio construction is described as combining multiple methodologies across markets so diversification and sizing, not only the individual signal rules, determine account outcomes. Fixed-fractional sizing is presented as a portfolio-level method that changes position size by market after several systems have been combined.

Value Charts are described as a normalized price display that marks moderately and significantly overbought or oversold conditions at plus or minus 4 and plus or minus 8, and that can be used across markets and time horizons.